US 12,103,418 B2
Electric vehicle charging optimization based on predictive analytics utilizing machine learning
Ronald J. Barber, San Jose, CA (US); Chad Eric DeLuca, Morgan Hill, CA (US); Rishabh Anup Nair, Pleasanton, CA (US); Uche Uba, Hawaiian Gardens, CA (US); Saisujit Madiraju, San Jose, CA (US); Niranjan Abhijeet Mirashi, San Jose, CA (US); Francisco Loya, Antioch, CA (US); and Emmanuel Shedu, Santa Clara, CA (US)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Aug. 20, 2020, as Appl. No. 16/998,833.
Prior Publication US 2022/0055496 A1, Feb. 24, 2022
Int. Cl. B60L 9/00 (2019.01); B60L 53/16 (2019.01); B60L 53/62 (2019.01); G06N 20/00 (2019.01); G08G 1/017 (2006.01); G08G 1/14 (2006.01); H02J 7/00 (2006.01)
CPC B60L 53/62 (2019.02) [B60L 53/16 (2019.02); G06N 20/00 (2019.01); G08G 1/017 (2013.01); G08G 1/143 (2013.01); H02J 7/0047 (2013.01)] 12 Claims
OG exemplary drawing
 
1. A method for managing charging resources of a charging system for plug-in electric vehicles (PEVs), the charging system including a central recording center comprising a tracking database, the method comprising:
initiating a charging session to a first PEV based on detecting that the first PEV has been plugged into a charging station;
associating the first PEV with the charging session in the tracking database; associating, in the tracking database, the first PEV to a first user and a first PEV profile;
charging the first PEV in accordance with information from the first PEV profile comprising identifying data and technical information needed for activating and managing charging operations;
monitoring and storing charging session data in the tracking database during the charging session;
tracking users based on responsiveness after notification of an available charging station, and feedback on other PEV owners;
rewarding some users with early notification of charging station availability and optimal waiting location identification;
generating a machine learning model based on collective charging data, wherein the machine learning model comprises a plurality of trained regression models including a linear regression model, a polynomial regression model and a K-nearest neighbor (KNN) model; and
predicting charge completion time of the first PEV based on an average of outputs from the linear regression model, the polynomial regression model, and the K-nearest neighbor (KNN) model;
wherein the identifying data comprises license plate number and manufacturer vehicle identification number, and the technical information comprises PEV: manufacturer, model, year, and battery size.